Recruitment Metrics That Matter: Hiring Funnel Guide

Recruitment Metrics That Matter: Hiring Funnel Guide

Most hiring teams at Indian SMBs track recruitment metrics the way people track their weight after a heavy meal: occasionally, anxiously, and with numbers they do not fully trust. Someone asks "how long is hiring taking?" in a leadership meeting, a recruiter opens a spreadsheet, and a figure gets quoted that nobody can reproduce next month. The problem is rarely effort. It is that the definitions are loose, the data is captured inconsistently, and the reporting ritual does not exist.

This guide fixes that. It walks through the recruitment metrics that genuinely change decisions, gives you a precise formula and data owner for each one, and shows you how to instrument your ATS so the numbers come out clean without anyone doing manual reconciliation on a Friday evening. You will also get a hiring funnel analytics framework, a weekly recruiting dashboard layout, a monthly review agenda, a stage-by-stage bottleneck diagnostic, a recruiter capacity model, and a 90-day rollout plan.

One promise up front: this article contains no borrowed industry benchmarks. You will not read that "the average time to hire is X days," because that number varies wildly by role, city, seniority and market conditions, and quoting it without your own data is worse than useless. Instead, you will learn how to build internal baselines you can defend.

Why Most Recruitment Metrics Fail Before They Start

Hiring data breaks for four boring reasons. Understanding them saves you months of chasing dashboards that lie.

Reason one: the definitions are ambiguous

Ask three people in your company when a role "opened" and you will get three answers. The day the founder said "we need a backend engineer." The day the budget was approved. The day the JD was published. Each answer produces a different time to fill, and the gap between them is often several weeks.

The same ambiguity infects every other metric. Is an "applicant" someone who clicked apply, or someone who submitted a complete profile? Is a referral that came through a job board still a referral? Without a written definition, you are averaging apples, oranges and guesswork.

Reason two: data entry is optional

If a recruiter can move a candidate from "screening" to "offer" without the intermediate stages being logged, they eventually will — especially under pressure. Every skipped stage silently corrupts your funnel conversion maths.

Reason three: the sample size is small

An SMB hiring 30 people a year has a genuinely small dataset. A single unusual hire can swing a monthly average by ten days. Small-sample volatility is not a reason to abandon measurement, but it is a reason to use medians, rolling windows and cohort views instead of single-month averages.

Reason four: nobody reviews the numbers

Metrics with no review ritual become decoration. If the dashboard is not opened in a recurring meeting where decisions get made, the data quality will decay within a quarter because there is no feedback loop punishing sloppiness.

The Twelve Recruitment Metrics That Actually Matter

Here is the core set. Resist the temptation to track more than this in year one. Each metric below has a precise formula, a clear owner and a stated decision it informs.

Add pipeline diversity as a thirteenth metric where relevant to your organisation — the definitions section below covers how to handle it responsibly.

Time to fill versus time to hire: stop conflating them

These two are constantly confused, and the confusion causes real arguments.

Time to fill measures the business's problem. It starts when the requisition is approved and ends when a candidate accepts. It includes all the time spent writing the JD, aligning on the scorecard, and waiting for the hiring manager to free up. It answers: "If I approve a role today, when will I have someone?"

Time to hire measures the recruiting process's efficiency for an individual candidate. It starts when a specific candidate enters your pipeline and ends when that candidate accepts. It answers: "Once we find the right person, how fast do we move?"

A role can have a 70-day time to fill and a 14-day time to hire. That combination tells a very specific story: sourcing is the bottleneck, not the interview process. The reverse — 30-day time to fill with a 26-day time to hire — says candidates are plentiful but your process is slow and you are probably losing people mid-funnel.

Track both. Report both. Never average them together.

Why medians beat averages in small teams

If eleven roles closed in 18 to 35 days and one niche role took 190 days, the average is dragged upward by a single outlier and misrepresents the typical experience. The median tells hiring managers what to expect; the outlier gets discussed separately as an exception.

Use this rule: report median as the headline, and report the 90th percentile alongside it so nobody forgets the tail. If you close fewer than eight roles a quarter, use a rolling six-month window instead of monthly cuts.

Defining Each Metric Precisely

Precision here is not pedantry. It is the difference between a number people act on and a number people argue about.

Time to fill

Start event: the date the requisition moves to "Approved" status in your system, with budget confirmed and a hiring manager assigned.

Stop event: the date the candidate confirms acceptance in writing.

Exclusions: requisitions cancelled or put on hold. If a role is put on hold for three weeks and reopened, either exclude the hold period from the clock or flag the requisition as "interrupted" — but pick one convention and write it down.

Common trap: back-dating requisition approval to when the manager first mentioned the need. This inflates the metric and creates arguments. Use the system timestamp only.

Time to hire

Start event: the date the eventually-hired candidate first entered your pipeline — application submitted, or first outreach message sent for a sourced candidate.

Stop event: offer acceptance date.

Nuance: for candidates who applied months ago, were rejected, and were revived for a new role, restart the clock at the revival date and tag the record as "re-engaged." Otherwise a strong revive looks like a 200-day disaster.

Cost per hire

The formula is simple; the arguments are about what goes inside it. Agree on the components before you compute anything.

Formula: Cost per hire = (Total internal recruiting cost + Total external recruiting cost) ÷ Number of hires in the same period.

Warning: cost per hire is a portfolio metric. Computing it for a single hire is almost always misleading, because fixed costs get dumped onto one head. Compute it monthly or quarterly across all hires, and separately by role family if you have enough volume.

Offer acceptance rate

Formula: (Offers accepted ÷ Offers released) × 100, counted by offer release date, not acceptance date.

Definition detail: count verbal offers as released only if you consistently do so. Many teams count only written offers, which hides the "verbal was declined so we never sent paper" cases. The cleaner approach is to log a verbal offer stage in the ATS and measure both verbal-to-written and written-to-accept.

Trap: renegotiated offers. If a candidate declines and then accepts a revised package, count one offer and one acceptance, and separately track a "renegotiation rate" so you can see how often your first number misses.

Source effectiveness

Source effectiveness has two halves and both matter.

• Source yield: hires from a source ÷ candidates entering from that source. Tells you which channels bring qualified people rather than volume.
• Source cost: total spend on the source ÷ hires from the source. Tells you what the channel costs per outcome.

A job board that generates 800 applications and two hires has a 0.25 percent yield. A referral channel with 40 candidates and six hires has a 15 percent yield. Volume without yield is noise that consumes recruiter hours.

Attribution rule: use first-touch attribution and write it down. If a candidate applies via a job board and is later referred, credit the job board. Multi-touch attribution is a rabbit hole for a team of your size.

Funnel conversion by stage

Define your stages once and never change them mid-quarter. A standard SMB funnel looks like this.

1. Applied or sourced (entered pipeline)
2. Recruiter screen completed
3. Hiring manager screen or assessment passed
4. Interview loop completed
5. Offer released
6. Offer accepted
7. Joined

Conversion is always forward-only: candidates entering stage N+1 ÷ candidates entering stage N. Withdrawals count as non-conversions but should be tagged separately so you can distinguish "we rejected them" from "they left us."

Interview-to-offer ratio

Formula: total candidates who completed the interview loop ÷ total offers released, in the same period, for the same role family.

A ratio of 3:1 means you interview three finalists per offer. A ratio of 12:1 means either your screening is not filtering, your interviewers are impossible to satisfy, or the role definition is unclear. All three are fixable, but only if you can see the number.

Quality of hire

Quality of hire is the metric everyone wants and few define. Do not try to make it perfect. Make it consistent.

Build a composite index from three to four inputs you can actually collect.

Normalise each input to a 0–100 scale, apply the weights, and you have a quality of hire score per cohort. Report it by hiring source, by hiring manager and by quarter of joining. That is where the insight lives — for example, discovering that referral hires score 78 while one particular job board's hires score 54.

Critical caveat: quality of hire is a lagging metric. A score for the January cohort is not complete until the following January. Report the 90-day partial score as a leading indicator and clearly label it as partial.

First-year attrition

Formula: (Employees who exited within 365 days of joining ÷ Employees who joined in the same cohort) × 100.

Use joining cohorts, not calendar periods. "Q1 FY26 joiner cohort: 22 joined, 4 exited within 12 months, 18.2 percent first-year attrition." That is interpretable. "First-year attrition in March was 8 percent" is not, because the denominator is undefined.

Split voluntary from involuntary exits. Early voluntary exits usually point at role misrepresentation, manager mismatch or a better counter-offer. Early involuntary exits point at screening failures. The fixes are completely different.

Candidate experience and NPS

Survey candidates at two points: after rejection at any stage, and after joining. Ask the standard 0–10 recommendation question plus two open text fields.

Formula: NPS = % promoters (9–10) − % detractors (0–6).

Segment by outcome. Rejected-candidate NPS and hired-candidate NPS are different animals and blending them hides everything useful. A healthy signal is rejected-candidate NPS that is not deeply negative — it means people felt respected even when the answer was no.

Keep response rate visible. An NPS built on a 6 percent response rate is a rumour, not a metric.

Diversity of pipeline

Measure representation at each funnel stage rather than only at the hire stage. The stage-by-stage view shows where representation drops, which is the only view that tells you what to fix.

Handle this data carefully: collect on a voluntary, self-declared basis, store it separately from evaluation records, report only in aggregate, and suppress any cell with fewer than five people to protect individual privacy. Follow your organisation's policies and applicable Indian data protection obligations, and involve your legal or compliance advisor before you start collecting.

Recruiter load

Raw requisition count per recruiter is a bad measure because a graduate hiring drive and a VP Engineering search are not comparable. Use a weighted model.

Recruiter load = sum of weights of open requisitions assigned. Set a ceiling per recruiter based on your own observed data, not on a number from the internet.

Instrumenting Your ATS So the Data Is Clean

Clean recruitment metrics are 80 percent instrumentation and 20 percent analysis. If you get this section right, the dashboards build themselves.

Step 1: Lock your stage model

Define a single stage list used by every requisition. Resist per-team customisation. If engineering wants an extra "take-home review" stage, add it as an optional sub-stage inside the assessment stage rather than a new top-level stage, so cross-role funnel comparison stays possible.

Write the entry criteria for each stage in one sentence. For example: "Recruiter screen complete = a 20–30 minute call happened and the screening scorecard was submitted."

Step 2: Make timestamps automatic, never typed

Every stage transition should be system-generated. Any date that a human types into a free-text field will eventually be wrong. If your ATS lets you disable manual date editing, disable it. If it does not, at least run an exception report on manually-edited dates each month.

Step 3: Make required fields genuinely required

The minimum required field set for every candidate record.

• Source (from a fixed dropdown, never free text)
• Source detail (which job board, which employee referred, which event)
• Requisition ID
• Rejection reason (from a fixed list) whenever a candidate is rejected
• Withdrawal reason whenever a candidate withdraws
• Offer details: date released, package band, date responded, accept/decline, decline reason

Step 4: Standardise your dropdown lists

Free text destroys analysis. "LinkedIn," "linkedin," "Linked In," and "LI" become four sources in your reports. Lock these lists.

Source list (example): Careers page, Employee referral, LinkedIn (organic), LinkedIn (paid), Naukri, Instahyre / niche board, Agency, Campus, Internal transfer, Rehire, Event or community, Other (requires a note).

Rejection reason list (example): Skill gap — technical, Skill gap — communication, Experience level mismatch, Compensation expectation mismatch, Location or work-mode mismatch, Notice period too long, Culture or values misalignment, Failed background verification, Position closed, Better candidate selected.

Decline reason list (example): Compensation, Counter-offer from current employer, Another offer accepted, Role or scope concerns, Location or work mode, Company stage or stability concerns, Personal reasons, No reason given.

That last list is the highest-value data in your entire ATS and the one most often left empty. Make it mandatory.

Step 5: Create the requisition record properly

The requisition, not the candidate, is the unit of business planning. Each requisition needs: unique ID, role title, level, department, hiring manager, location, work mode, budget band, approval date, target join date, priority tier, and status history (open / on hold / cancelled / filled).

Step 6: Set up an automated data hygiene check

Run a weekly exception report. If your system supports scheduled reports, automate it and send it to the TA lead every Monday morning.

Step 7: Connect the ATS to the HRIS

The handoff from "offer accepted" to "employee joined" is where most SMB data pipelines break. If your ATS and HRIS are separate systems with a manual copy-paste in between, you will lose the link between the hire record and the employee record — and without that link, quality of hire and first-year attrition are impossible to compute.

An integrated HRMS platform such as CozyHR removes this seam by carrying the candidate ID through to the employee record, so joining dates, probation outcomes and exit dates flow back into your hiring analysis automatically.

Building the Hiring Funnel: A Worked Illustrative Example

Everything below is illustrative arithmetic for a fictional company. These are not benchmarks, not survey data, and not typical figures for any industry. They exist to show the method.

Meet Lattice Foods Pvt Ltd (fictional), a 140-person D2C food brand in Pune hiring across engineering, sales and operations. Here is one quarter of funnel data for their Sales Executive role family.

The funnel conversion worksheet

Reading this illustrative funnel:

• Top-of-funnel volume is high but only 18.8 percent survive the recruiter screen. That is a targeting problem: the job ad or the channel mix is attracting people who do not match.
• The interview loop converts at 47.1 percent, giving an interview-to-offer ratio of roughly 2.1:1 (34 ÷ 16). That is reasonably tight.
• The offer-to-accept rate is 75 percent and then two of twelve accepted candidates did not join — a 16.7 percent post-acceptance drop, and a 32-day median offer-to-join gap. That gap is where the losses are happening.

Working out time to fill and time to hire

Lattice Foods closed six Sales Executive requisitions in the quarter. Illustrative figures.

Median time to fill: sort the values — 26, 27, 36, 48, 50, 52. The median of the middle two (36 and 48) is 42 days.

Median time to hire: sorted — 18, 19, 19, 20, 25, 27. Median of 19 and 20 is 19.5 days.

Median time to start: sorted — 29, 30, 32, 34, 34, 35. Median of 32 and 34 is 33 days.

The diagnosis writes itself. Time to hire is 19.5 days but time to fill is 42 days. The 22-day difference is time spent before the eventual hire ever entered the pipeline — that is sourcing lag, not process lag. Speeding up interviews would save almost nothing. Building a pre-qualified sales talent pool would cut roughly three weeks off every future requisition.

Meanwhile, the 33-day time to start combined with two no-joins says the offer-to-join window needs an engagement programme.

Working out cost per hire

Illustrative quarterly recruiting costs for Lattice Foods across all 22 hires in the quarter (not just sales).

Blended cost per hire = ₹21,90,000 ÷ 22 = ₹99,545 per hire (illustrative).

That blended number is not very actionable on its own. Split it.

Shared cost here is the ₹13,50,000 of recruiter salaries, tools, verification and interviewer time allocated at ₹63,000 per hire.

What this illustrative split reveals: agency hires cost roughly three times a referral hire. If the referral channel could be scaled — even by doubling the bonus — the economics would still favour it heavily. That is a decision the blended number would never have surfaced.

Working out source effectiveness

Illustrative quarter, all roles.

Reading this illustrative table: Job board B consumes 390 candidate reviews to produce two hires with the lowest quality score and the highest non-agency cost. That is a clear cut-or-renegotiate decision. The 390 screened candidates also represent enormous hidden recruiter time — roughly 65 hours at 10 minutes each — which is the real cost.

Note how source effectiveness only becomes decision-grade when yield, cost, speed and quality sit in the same table. Any one column alone would mislead you.

Building Your Recruiting Dashboard

A recruiting dashboard is not a data dump. It is a weekly decision aid. Keep it to one screen.

The weekly dashboard layout

Panel 8 is non-negotiable and belongs on the same screen as everything else. When a data hygiene exception count is visible to the whole team every week, it drops fast.

Leading versus lagging indicators

Split your dashboard mentally into two halves.

Leading indicators — you can act on these this week: - New candidates entering the pipeline per open requisition - Screens completed per recruiter per week - Interviews scheduled for the coming week - Candidates stalled more than seven days - Interview feedback submitted within 24 hours (percentage)

Lagging indicators — these confirm whether past decisions worked: - Time to fill - Cost per hire - Offer acceptance rate - Quality of hire - First-year attrition

A common failure is a dashboard made entirely of lagging indicators. It tells you the quarter went badly, three weeks after you could have done anything.

Making the dashboard actually get used

• Put it in one place, not five spreadsheets.
• Have it auto-refresh so nobody is manually rebuilding it.
• Show week-on-week deltas, not just absolute numbers.
• Highlight breaches of your own thresholds in colour.
• Never add a metric without deleting one, after the first six months.

The Monthly Hiring Review Ritual

The dashboard is the instrument. The review is the engine.

Weekly stand-up: 20 minutes, recruiters plus TA lead

Fixed agenda:

1. Requisitions where the plan is at risk (aging or zero-pipeline reqs).
2. Candidates stalled more than seven days — who unblocks each one, by when.
3. Offers pending response, with the closing plan for each.
4. Data hygiene exceptions from Monday's report, cleared before Friday.
5. One process irritant to fix this week.

Monthly hiring review: 60 minutes, TA lead plus hiring managers plus finance

Fixed agenda with time boxes:

1. Plan versus actual (10 min). Hires planned, hires made, hires forecast for next month. Gap explanation.
2. Speed (10 min). Median time to fill and time to hire, with the split between sourcing lag and process lag. Which stage moved.
3. Funnel conversion (10 min). Stage conversions versus the prior three-month baseline. One stage chosen for focused improvement.
4. Offers and declines (10 min). Acceptance rate and every decline reason read aloud. Compensation band pressure flagged to finance.
5. Cost and source mix (10 min). Cost per hire by segment, channel decisions for next month.
6. Quality signals (5 min). 90-day quality scores for the cohort that just crossed 90 days, and any early attrition.
7. Decisions and owners (5 min). Every item leaves with a name and a date.

Quarterly deep dive: 90 minutes, add the leadership team

• Baseline refresh: recompute your internal benchmarks from the last 12 months of data.
• Capacity plan for the next quarter against the hiring forecast.
• Quality of hire by source and by hiring manager, with a full year of data where available.
• Interviewer calibration review: who is systematically harsher or softer than the panel.
• Requisition process review: how long approvals actually take.

The one rule that makes reviews work

Every metric that moves in the wrong direction gets one named owner and one specific action with a date. Not "we should improve screening." Instead: "Priya rewrites the sales JD and screening questions by the 18th; we review screen-pass rate in the next monthly."

Diagnosing Bottlenecks Stage by Stage

This is the most practically useful table in this guide. Find your symptom, read across.

How to work the table properly

Fix one stage at a time. If you change the JD, the channel mix and the interview loop in the same month, you will never know which change worked — and with SMB sample sizes you may not be able to detect it statistically at all.

Give each change a full hiring cycle before judging it. For most SMB roles that is six to ten weeks.

Capacity Planning for Recruiters

The question "do we need another recruiter?" should be answered with arithmetic, not vibes.

Step 1: Establish your own throughput baseline

Look back over the last six months and compute, per recruiter:

• Hires completed
• Weighted requisitions carried on average
• Screens conducted
• Interviews coordinated
• Offers released

Do not import a number from an article. Your process, your market and your role mix determine your throughput.

Step 2: Build the demand forecast

Get the hiring plan for the next two quarters from finance and department heads. Convert each planned hire into a weighted requisition using the difficulty weights defined earlier.

Illustrative example — Lattice Foods, next quarter:

Step 3: Convert demand into recruiter months

Assume — from your own baseline, illustratively — that one recruiter can carry 8 weighted requisitions concurrently and close roughly 4 weighted requisitions per month at that load.

Weighted demand of 27.0 over a three-month quarter requires 27.0 ÷ 3 = 9.0 weighted closures per month. At 4 per recruiter per month, that is 2.25 recruiters needed.

If you have two recruiters, you are 0.25 short — about 11 percent over capacity. Options in order of cost:

1. Cut low-yield sourcing work to free recruiter hours (cheapest, often the biggest win).
2. Push non-urgent requisitions into the following quarter.
3. Use an agency for the single leadership role, which consumes 3 of the 27 weighted points.
4. Hire a sourcing associate rather than a full recruiter.
5. Hire a third recruiter.

Step 4: Account for the hidden load

Recruiters do not spend 100 percent of their time on requisitions. Budget realistically for:

• Interview coordination and rescheduling
• Offer paperwork and background verification follow-ups
• Hiring manager alignment meetings
• Candidate experience work: rejections, follow-ups, referral nurturing
• Data hygiene and reporting
• Their own leave

If you plan capacity assuming full-time requisition work, you will be permanently over capacity and will not understand why.

Step 5: Watch the leading signals of overload

• Median days in the recruiter screen stage rising
• Stalled candidate count climbing week on week
• Data hygiene exceptions increasing
• Time to first qualified candidate stretching
• Recruiter-driven sourcing dropping while inbound-only reliance rises

These appear weeks before time to fill degrades. Treat them as your early warning system.

Common Mistakes and Vanity Metrics

Some numbers look like insight and are not.

Vanity metric: total applications received

Application volume measures advertising reach, not hiring health. A campaign that triples applications while halving screen-pass rate has made things worse — it has added screening hours without adding hires.

Track instead: qualified candidates per requisition, and screen-pass rate by source.

Vanity metric: number of interviews conducted

Interview volume is an input cost, not an output. High interview counts with low offer counts mean your screening is failing and you are burning your team's calendar.

Track instead: interview-to-offer ratio, and interview hours per hire.

Vanity metric: average time to fill across all roles

Averaging a warehouse role and a VP search produces a number that describes neither. Worse, it makes the metric look like it is improving whenever you happen to close easy roles.

Track instead: median time to fill segmented by role family and level, with the 90th percentile shown alongside.

Vanity metric: cost per hire as a target to minimise

Cost per hire is a diagnostic, not a goal. You can drive it to near zero by only hiring through free channels — and quietly destroy quality of hire and time to fill in the process.

Track instead: cost per hire alongside quality of hire and time to fill. Judge them as a set.

Mistake: comparing yourself to unverified external benchmarks

You will find plenty of confident claims online about what "average" time to hire, cost per hire or offer acceptance rate should be. Treat them with deep suspicion. Most are drawn from unstated samples, different definitions, different countries, different role mixes and different economic conditions. A number computed from US enterprise tech hiring tells a 60-person Coimbatore manufacturing SMB nothing at all.

Do this instead — build internal baselines:

1. Pull the last 12 months of hiring data, cleaned.
2. Segment by role family and level.
3. Compute median and 90th percentile for each time-based metric within each segment.
4. Compute conversion rates per stage per segment.
5. Write these down as your baseline, dated.
6. Compare every future period against your own baseline, not someone else's.
7. Refresh the baseline every six months, keeping the old version for trend comparison.

If you have less than a year of clean data, say so explicitly on the dashboard: "Baseline provisional — 4 months of data, 11 hires." Honesty about sample size protects you from over-interpreting noise.

Mistake: changing definitions mid-stream

The moment you redefine time to fill, your entire history becomes incomparable. If you must change a definition, recompute the historical series under the new definition, or clearly mark the break point on every chart.

Mistake: measuring recruiters on metrics they do not control

A recruiter cannot control a hiring manager who takes nine days to give feedback, or a compensation band set below market. Holding them accountable for time to fill alone creates gaming: candidates get moved to "offer" prematurely, or reqs get quietly reopened to reset the clock.

Assign accountability by control:

Mistake: reporting a percentage on a tiny denominator

"Offer acceptance rate fell to 50 percent this month" sounds alarming until you learn it was one decline out of two offers. Always publish the numerator and denominator next to any percentage. Suppress percentages entirely below a denominator of five.

Mistake: ignoring withdrawal versus rejection

If a candidate leaves your process, it matters enormously whether you rejected them or they walked away. Blending both into "did not convert" hides your single most important signal about candidate experience and process speed.

Mistake: treating the dashboard as the deliverable

The dashboard is not the outcome. The outcome is a decision: cut this channel, fix this stage, add this recruiter, raise this band. If a month passes with no decision made from the data, the measurement programme is not working.

A 90-Day Rollout Plan

You do not need a data team. You need sequencing and discipline.

Days 1–15: Define and align

Goal: everyone agrees what the words mean.

1. Write one-page definitions for all twelve core metrics, including start events, stop events and exclusions.
2. Lock the stage model and write entry criteria for each stage.
3. Finalise the dropdown lists: sources, rejection reasons, decline reasons.
4. Agree the cost per hire component list with finance in writing.
5. Set the requisition record template with all mandatory fields.
6. Circulate the definitions document to hiring managers and get explicit sign-off.

Deliverable: a metric definitions document that anyone can read and apply. This is the single highest-leverage artefact in the whole programme.

Days 16–30: Instrument the system

Goal: the data starts capturing itself.

1. Configure the stage model in the ATS.
2. Make required fields mandatory and disable manual date editing where possible.
3. Replace all free-text sources with the locked dropdown.
4. Build the weekly data hygiene exception report.
5. Establish the ATS-to-HRIS link so hire records connect to employee records.
6. Run a 45-minute training session for recruiters and hiring managers on the new fields and why they matter.

Deliverable: a configured ATS producing clean data from day 30 onward.

Days 31–45: Clean history and build baselines

Goal: know where you actually stand.

1. Export the last 12 months of hiring data.
2. Clean it: deduplicate candidates, fill missing sources where recoverable, mark unrecoverable records as "unknown" rather than guessing.
3. Compute baselines for each metric by role family: median, 90th percentile, stage conversions.
4. Document explicitly what is missing and how confident you are in each baseline.
5. Present the baseline to leadership with clear caveats about sample size.

Deliverable: a dated internal baseline document. Do not skip the caveats — they protect the programme's credibility.

Days 46–60: Build the dashboard

Goal: one screen, refreshed automatically.

1. Build panels 1 through 5 first (requisitions, funnel, speed, conversion, offers). These deliver the most value fastest.
2. Add the data hygiene panel — it goes live with the first five, not later.
3. Set up automatic refresh. Manual rebuilds die within six weeks.
4. Add threshold highlighting against your own baselines.
5. Run it in parallel with existing reporting for two weeks and reconcile any differences.

Deliverable: a live weekly recruiting dashboard that nobody has to rebuild.

Days 61–75: Start the rituals

Goal: the data starts driving decisions.

1. Launch the 20-minute weekly stand-up with the fixed agenda.
2. Run the first monthly hiring review with hiring managers and finance.
3. Log every decision with an owner and a date.
4. Track hygiene exception counts weekly and drive them toward zero.
5. Pick exactly one funnel stage to improve and run a focused experiment.

Deliverable: two consecutive weeks of stand-ups and one completed monthly review with logged decisions.

Days 76–90: Add quality and close the loop

Goal: connect hiring decisions to hiring outcomes.

1. Launch the 90-day hiring manager satisfaction survey.
2. Build the quality of hire composite and compute it for the earliest available cohort.
3. Launch candidate experience surveys for rejected and hired candidates.
4. Set up first-year attrition cohort tracking in the HRIS.
5. Run the first quarterly deep dive, refresh baselines, and remove any metric nobody has used.

Deliverable: quality and retention signals flowing back into source and process decisions.

What to deliberately not do in the first 90 days

• Do not build predictive models. You do not have the data volume.
• Do not track more than the core twelve metrics.
• Do not benchmark against external numbers.
• Do not attempt multi-touch source attribution.
• Do not automate anything you have not done manually at least twice.

Adapting This for Indian SMB Realities

A few practical adjustments that matter in the Indian hiring context.

Notice periods change the shape of everything

Long notice periods, common in many Indian sectors, mean time to start can exceed time to hire by a wide margin. Two consequences:

• Report time to start as a first-class metric, not a footnote. Your business plan depends on when someone actually starts, not when they say yes.
• Instrument the offer-to-join window deliberately: log notice period at the screening stage, track buy-out requests, and record every no-join reason.

The counter-offer dynamic

Counter-offers from current employers are a frequent cause of both declines and no-joins. Capture "counter-offer from current employer" as an explicit decline reason and as a no-join reason. If that reason clusters in one role family, your compensation bands need review, not your recruiters.

Multi-location and hybrid complexity

If you hire across cities with different market rates, segment your metrics by location. A single blended offer acceptance rate across Bengaluru, Indore and Kochi hides the location where your band is actually uncompetitive.

Volume hiring alongside specialist hiring

Many Indian SMBs run both. Never blend them in a single report. Volume hiring is measured on throughput, cost per hire and 90-day retention. Specialist hiring is measured on time to fill, quality of hire and offer acceptance rate. Different funnels, different dashboards, different recruiter skill sets.

Referral culture

Referral channels frequently show the strongest yield and quality in Indian SMBs — but only if you instrument them. Log the referring employee, track referral hire quality separately, and measure how long referral candidates wait for a first response. A referral that gets no reply for ten days damages two relationships at once.

Frequently Asked Questions

What is the difference between time to fill and time to hire?

Time to fill starts when the requisition is approved and ends when a candidate accepts the offer — it measures the business's total wait. Time to hire starts when the eventually-hired candidate first entered your pipeline and ends at acceptance — it measures how fast you move a specific candidate through your process. The gap between the two is your sourcing lag. If time to fill is 45 days and time to hire is 15 days, 30 days were spent before the right candidate ever appeared, which means the fix is sourcing, not interview scheduling.

How many recruitment metrics should a small company track?

Start with six: median time to fill, median time to hire, offer acceptance rate, stage conversion rates, cost per hire and source yield. Add quality of hire and first-year attrition once you have a year of connected data. Twelve is a reasonable ceiling for most SMBs. Tracking thirty metrics with a two-person team means tracking none of them well, and every additional metric adds data entry burden that degrades the quality of the ones that matter.

How do I calculate cost per hire when recruiters work on many roles at once?

Do not attempt to attribute recruiter salary to individual hires. Compute cost per hire at the portfolio level: total recruiting cost for a period divided by hires in that period. If you need role-level insight, split directly attributable costs (agency fees, referral bonuses, role-specific ads) to the specific hire, then allocate shared costs evenly across all hires in the period. Document your allocation method and never change it mid-year, because changing the method makes trend comparison meaningless.

What is a good time to hire for my industry?

There is no reliable universal answer, and any article giving you a specific number for "your industry" is almost certainly quoting an unverifiable figure derived from a different market, a different role mix and a different definition. Build your own baseline instead: compute median time to hire from your last 12 months, segmented by role family and level, and use that as your reference point. Your goal is beating your own previous quarter, not matching a statistic of unknown provenance.

How do I measure quality of hire without a formal performance system?

Use a lightweight composite that does not require performance software. Send the hiring manager a three-question survey at day 90: would you hire this person again (yes/no), how is their ramp progressing against expectations (ahead/on time/behind), and how would you rate them overall (1–5). Combine those with a simple retained-at-12-months flag from your HRIS. That gives you a defensible quality signal you can segment by source and by hiring manager, which is where the actionable insight lives.

What should I do if my offer acceptance rate is dropping?

Read the decline reasons before changing anything — that is the entire diagnosis. If declines cluster on compensation, your bands need a market review, and you should pre-close on numbers during the recruiter screen so you never release an offer into a known gap. If declines cluster on "accepted another offer," you are too slow: measure the days from final interview to offer release and compress it below 72 hours. If declines cluster on role or scope concerns, your JD and interview messaging are misrepresenting the job. Never make all three changes at once, because you will not know which one worked.

How do I get hiring managers to actually enter data in the ATS?

Reduce what you ask them for, then make the value visible. Hiring managers typically need to do exactly two things: submit interview feedback with a rating within 24 hours, and select a rejection reason from a dropdown. That is it — everything else should be the recruiter's job. Then show them, in the monthly review, how their own data answered a question they cared about. When a manager sees that their feedback turnaround was the reason two candidates were lost, compliance stops being a compliance problem.

Should I track diversity metrics in my hiring funnel?

If diversity is a stated organisational goal, measure it at each funnel stage rather than only at the hire — stage-level data shows you where representation drops, which is the only view that suggests a fix. Collect the data voluntarily and self-declared, store it separately from evaluation records, report only in aggregate, and suppress any group smaller than five people to protect individual privacy. Consult your legal or compliance advisor before you begin collecting, and make sure the data is used to examine your process, never to influence an individual hiring decision.

Bringing It Together

Recruitment metrics are not a reporting exercise. They are a way of turning a set of contested opinions about hiring into a shared set of facts that a small team can act on quickly.

The sequence matters more than the sophistication. Define your metrics precisely. Instrument the ATS so the data captures itself. Build internal baselines instead of importing benchmarks you cannot verify. Put one dashboard in front of the team every week and run one review every month where decisions get made and owners get named. Then fix one stage at a time and give each change a full hiring cycle to prove itself.

Do that for two quarters and you will be able to answer the questions that actually matter: how long will this role take, what will it cost, which channel is worth the money, where is the funnel leaking, and are the people we hire working out. Those are the answers that let a growing company plan.

If your hiring data currently lives across an ATS, three spreadsheets and a WhatsApp group, the single biggest improvement available to you is a connected system where the requisition, the candidate, the offer and the employee record are the same thread. CozyHR brings recruitment, onboarding, payroll and HR operations into one platform built for Indian SMBs, so your hiring funnel analytics and your people data stop living in different worlds. Try CozyHR and see what your hiring numbers look like when they finally reconcile.